Walrus: Erasure coding basics, why blobs survive node failures on Sui
I’ve been trying to understand why storage can stay reliable even when some machines go offline. Walrus uses erasure coding: when you upload a blob, it’s split into many small pieces plus extra “repair” pieces, then spread across many nodes. To read it back, the network doesn’t need every piece just enough pieces to reconstruct the original data so a few node failures or missed replies don’t automatically break retrieval.It’s like tearing a document into many strips, making a few spare strips, and only needing most of them to reassemble it.fees pay for writes/reads, staking secures storage operators, and governance tunes parameters like redundancy and penalties.I’m not fully sure how this feels under extreme demand spikes, since real-world latency is hard to predict. #Walrus @Walrus 🦭/acc $WAL
